Email list hosting service & mailing list manager

Libraries at scheme.org? Lassi Kortela (26 Jan 2021 22:49 UTC)
Re: Libraries at scheme.org? Vladimir Nikishkin (27 Jan 2021 01:28 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 07:28 UTC)
Re: Libraries at scheme.org? Vladimir Nikishkin (27 Jan 2021 01:34 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 07:46 UTC)
Re: Libraries at scheme.org? Arvydas Silanskas (27 Jan 2021 08:26 UTC)
Re: Libraries at scheme.org? Marc Nieper-Wißkirchen (27 Jan 2021 08:36 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 08:51 UTC)
Re: Libraries at scheme.org? Arvydas Silanskas (27 Jan 2021 10:34 UTC)
Re: Libraries at scheme.org? Marc Feeley (27 Jan 2021 13:17 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 14:07 UTC)
Re: Libraries at scheme.org? Marc Feeley (27 Jan 2021 16:28 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 17:34 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 18:10 UTC)
Re: Libraries at scheme.org? Marc Feeley (27 Jan 2021 19:54 UTC)
Re: Libraries at scheme.org? Lassi Kortela (29 Jan 2021 13:28 UTC)
Re: Libraries at scheme.org? Lassi Kortela (27 Jan 2021 08:37 UTC)
Re: Libraries at scheme.org? Duy Nguyen (28 Jan 2021 10:29 UTC)

Re: Libraries at scheme.org? Lassi Kortela 27 Jan 2021 14:07 UTC

> One way to achieve this is with a syntax for module/libraries that includes an (optional) version number, combined with the ability to create aliases.  So if B says
>
>    (import (C))
>
> you could
>
>    (define-module-alias (C) (C @v1.0.1))
>
> in a global configuration file to force the import in B to use v1.0.1 of C.

Good idea.

Are you interesting in standardizing the format of such a configuration
file? I think we should, and been thinking about it on and off for about
a year. The snow-fort.org package.scm files are a reasonable starting
point. Chicken egg files contain somewhat similar information.

The library name (C @v1.0.1) is almost equivalent to the R6RS-style
library name (C (1 0 1)) or (C (v1.0.1)).

IMHO the optimum solution would be to:

- extend the R7RS rules to permit a list as the last library name part
to give the version number

- extend the R6RS rules to permit non-negative exact integers as library
name parts; and to permit symbols in addition to integers in the version
number part